Spent the weekend chasing a random seal failure on a hot oil pump. HMI looked green. The logs showed NPSHa collapsing whenever the upstream strainer loaded after CIP, then a hard restart with the VFD ramping like a dragster. Three seconds of dry run at 1,800 rpm. Well, there’s your problem.
We keep hoping operators will watch suction, level, and valve positions. That is not control. Interlock the start: no run unless suction pressure > X and rising, tank level > Y, and valve open proven. Validate the permissives for 5 s. Trip on fast negative dP trend across the strainer. Tie ramp rate to NPSH margin, not a fixed time.
Expensive? Less than a seal every quarter and a cooked sleeve. Also quiets the HMI. One clear Start permissive failed: suction low beats a dozen blinking faults. If you are writing a new SOP for a chronic pump trip, stop. Put it in logic.
What single permissive or trip saved you the most grief on rotating equipment? Anyone using clever cavitation proxies beyond suction transmitters and vibration?
Biggest saver: no start unless min‑flow recirc proven open with flow >= Qmin, plus a 2 s “no discharge pressure rise” trip. For a cavitation proxy we watch VFD torque/current ripple (1 s stdev) and a cheap piezo on the casing; both spike before suction P alarms so we slow the ramp or block start. After CIP we freeze starts until strainer dP trend is flat for 5 s. Just interlock it.